Vision Foundation Model Enables Generalizable Object Pose Estimation